1

I'm mostly interested in C++ answers for this but in addition C answers would be welcome too as they will probably vary. This is a simplified version of the real code to focus on my question so assume what I've written is correct :P

Imagine I have a function called process in a library.
This function takes three parameters, a, b, and c
Most of the time the function only needs the values of a and b to do it's job but in some circumstances requires c.

Now I want to call this from a main program.
process(1, 2, very_expensive_call(55, 67, "Hello"));

Imagine that the value of c is very expensive to calculate. As it's probably not needed I don't want to calculate it unless I have to.

I don't want my main program to have to repeat the logic of knowing if c will be needed. I don't want the library to know about very_expensive_call.

So how can I organize my code to prevent the call to the expensive function while remaining both efficient and elegant?

too honest for this site
  • 12,050
  • 4
  • 30
  • 52
jcoder
  • 29,554
  • 19
  • 87
  • 130

1 Answers1

2

Make it a lazy call, pass a function object that will calculate the value on request:

template<typename Expensive>
int process(int, int, Expensive&&);

and then you'd call:

process(1, 2, []{ return very_expensive_call(55, 67, "Hello"); });

If you do not want to make your function a template, use std::function for type erasure (or possibly a lighter alternative if there is one that suits your needs):

int process(int, int, function<expensive_result()>);

with no change at the call site:

process(1, 2, []{ return very_expensive_call(55, 67, "Hello"); });

This solution will require you to explicitly call the operator () in the process's code to get compute the value.

krzaq
  • 16,240
  • 4
  • 46
  • 61